Dhruv Nair

@rafationgson can you check to see if the individual frames of the video are blank as well?

Hi @lulu1315 thanks for flagging. I've opened a PR to support converting AnimateDiff safetensors format files to Diffusers.

Don't think the timeout should be hardcoded. I think it's better to define a constant e.g `DIFFUSERS_REQUEST_TIMEOUT` that you should be able to override via environment variable and then pass...

> I also think we do not need to accept list of 5d (the video processor should only accept video or a list of videos, and 5d is already a...

Hi @liyy201912 What do you mean by incorrect generation here? Do you have an example output?

Hmm the results do look good. But from the project page they mention > We leverage methods from stochastic calculus and find optimal schedules specific to different solvers, trained DMs...

> I'm a little bit confused, it says each scheduler would have its own optimized schedule but this is what they provide. Can these timesteps be used for all schedulers...

cc: @yiyixuxu and @asomoza for visibility.

cc @yiyixuxu could you take a look here please.

From your logs it looks like you tried to run inference with the pipeline? Can you share a full code example of exactly what you are running that is causing...